linux-kernel - Re: [PATCH v2 1/6] locking/lockdep: Track number of zapped classes

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <2e9d25b1-f61a-2367-36f3-7b39a9cca095@acm.org>
Date:   Mon, 13 Jan 2020 08:02:11 -0800
From:   Bart Van Assche <bvanassche@....org>
To:     Waiman Long <longman@...hat.com>,
        Peter Zijlstra <peterz@...radead.org>
Cc:     Ingo Molnar <mingo@...hat.com>, Will Deacon <will.deacon@....com>,
        linux-kernel@...r.kernel.org
Subject: Re: [PATCH v2 1/6] locking/lockdep: Track number of zapped classes

On 1/13/20 6:58 AM, Waiman Long wrote:
> On 1/13/20 9:55 AM, Peter Zijlstra wrote:
>> On Mon, Dec 16, 2019 at 10:15:12AM -0500, Waiman Long wrote:
>>> The whole point of the lockdep dynamic key patch is to allow unused
>>> locks to be removed from the lockdep data buffers so that existing
>>> buffer space can be reused. However, there is no way to find out how
>>> many unused locks are zapped and so we don't know if the zapping process
>>> is working properly.
>>>
>>> Add a new nr_zapped_classes variable to track that and show it in
>>> /proc/lockdep_stats if it is non-zero.
>>>
>>> diff --git a/kernel/locking/lockdep_proc.c b/kernel/locking/lockdep_proc.c
>>> index dadb7b7fba37..d98d349bb648 100644
>>> --- a/kernel/locking/lockdep_proc.c
>>> +++ b/kernel/locking/lockdep_proc.c
>>> @@ -336,6 +336,15 @@ static int lockdep_stats_show(struct seq_file *m, void *v)
>>>   	seq_printf(m, " debug_locks:                   %11u\n",
>>>   			debug_locks);
>>>   
>>> +	/*
>>> +	 * Zappped classes and lockdep data buffers reuse statistics.
>>> +	 */
>>> +	if (!nr_zapped_classes)
>>> +		return 0;
>>> +
>>> +	seq_puts(m, "\n");
>>> +	seq_printf(m, " zapped classes:                %11lu\n",
>>> +			nr_zapped_classes);
>>>   	return 0;
>>>   }
>> Why is that conditional?
>>
> Because I thought zapping class doesn't occur that often. Apparently,
> class zapping happens when the system first boots up. I guess that
> conditional check isn't needed. I can remove it in the next version.

Zapping a lock class happens every time a lock key is unregistered. That 
can happen any time as one can see in the following grep output:

$ git grep -lw 'lockdep_unregister_key'
drivers/net/bonding/bond_main.c:4446: 
lockdep_unregister_key(&bond->stats_lock_key);
drivers/net/team/team.c:1676:	lockdep_unregister_key(&team->team_lock_key);
drivers/net/team/team.c:1984:		lockdep_unregister_key(&team->team_lock_key);
include/linux/lockdep.h:294:extern void lockdep_unregister_key(struct 
lock_class_key *key);
include/linux/lockdep.h:464:static inline void 
lockdep_unregister_key(struct lock_class_key *key)
kernel/locking/lockdep.c:5166:void lockdep_unregister_key(struct 
lock_class_key *key)
kernel/locking/lockdep.c:5201:EXPORT_SYMBOL_GPL(lockdep_unregister_key);
kernel/workqueue.c:3456:	lockdep_unregister_key(&wq->key);
net/core/dev.c:9172:	lockdep_unregister_key(&dev->qdisc_tx_busylock_key);
net/core/dev.c:9173:	lockdep_unregister_key(&dev->qdisc_running_key);
net/core/dev.c:9174:	lockdep_unregister_key(&dev->qdisc_xmit_lock_key);
net/core/dev.c:9175:	lockdep_unregister_key(&dev->addr_list_lock_key);
net/core/dev.c:9183:	lockdep_unregister_key(&dev->qdisc_xmit_lock_key);
net/core/dev.c:9184:	lockdep_unregister_key(&dev->addr_list_lock_key);
tools/lib/lockdep/include/liblockdep/common.h:48:void 
lockdep_unregister_key(struct lock_class_key *key);
tools/lib/lockdep/include/liblockdep/mutex.h:58: 
lockdep_unregister_key(&lock->key);

Bart.